Production of structured triacylglycerols in an immobilised lipase packed-bed reactor: batch mode operation

Transcription

1 Journal of Chemical Technology and Biotechnology J Chem Technol Biotechnol 80: ) DOI: /jctb.1149 Production of structured triacylglycerols in an immobilised lipase packed-bed reactor: batch mode operation PA González Moreno, 1 A Robles Medina, 1 F Camacho Rubio, 2 B Camacho Páez, 1 L Esteban Cerdán 1 and E Molina Grima 1 1 Departamento de Ingeniería Química, Universidad de Almería, Spain 2 Departamento de Ingeniería Química, Universidad de Granada, Spain Abstract: Structured triacylglycerols with caprylic acid at the sn-1 and sn-3 positions of the glycerol backbone and eicosapentaenoic acid EPA) at the position sn-2 were synthesised by acidolysis of a commercially available EPA-rich oil EPAX4510, Pronova Biocare) and caprylic acid catalysed by the 1,3-specific immobilised lipase Lipozyme IM. The reaction was carried out in an immobilised lipase packed-bed reactor by recirculating the reaction mixture through the bed. The exchange equilibrium constants between caprylic acid and the native fatty acids of EPAX4510 were determined. The n-3 polyunsaturated fatty acids PUFAs), EPA and docosohexaenoic acid DHA), were the most easily displaced by the caprylic acid. The exchange equilibrium constants were 3.68 and 3.06 for EPA and DHA, respectively. The influence of the flow rate of the reaction mixture through the packed-bed and the substrate concentration in the reaction rate were studied. For flow rates between 74 and 196 cm 3 h 1 bed of 6.6 mm internal diameter and 0.46 porosity) and triacylglycerol concentrations between and M, the data fitted well to an empirical kinetic model which allowed representative values of the apparent kinetic constant to be obtained.hence,the average reaction rates and kinetic constants of exchange of caprylic acid and native fatty acids of EPAX4510 could be calculated. In the conditions indicated, the parameter lipase mass time/triacylglycerol mass, m L t/v[tg] 0 ) constituted the intensive variable of the process for use in predicting the composition of structured triacylglycerols at different reaction times. At equilibrium, the structured triacylglycerol produced had the following composition: caprylic acid 59.5%, EPA 9.6%, DHA 2.2% and oleic acid 11.8% Society of Chemical Industry Keywords: lipase; kinetics; structured triacylglycerols; packed-bed reactor; recirculation; acidolysis interesterification); eicosapentaenoic acid EPA); polyunsaturated fatty acid; caprylic acid 1 INTRODUCTION For nutritional purposes, there is an increasing interest in the production of structured triacylglycerols ST) containing medium chain fatty acids M) located at the positions sn-1 and sn-3 of the glycerol backbone and functional long chain polyunsaturated fatty acids L) located at the position sn-2 MLM). These structured triacylglycerols are claimed to impart beneficial effects on immune function and help improve lipid clearance from the bloodstream. 1 STs with eicosapentaenoic acid EPA) and docosohexaenoic acid DHA) predominantly in the position sn-2 are readily absorbed compared with other EPA/DHA-containing STs. In general, STs are promising for both enteral and parenteral nutrition. 2 There is clinical evidence that the MLM structure of structured triacylglycerols improves the absorption of fat. Christensen et al 3 have compared the absorption of EPA and DHA by intragastric administration of two oils with different intramolecular triacylglycerol structures one with EPA and DHA located in the sn-2 position, the other with a random fatty acid distribution). The first was the more readily absorbed source of EPA and DHA. 3 Kennedy et al 4 have proved the advantages of administrating the palmitate acyl-group in the position sn-2 of the glycerol backbone and this author has also proved the following advantages: higher whole-body bone mineral content, reduced stool soap fatty acids, and softer stools, more like those of breast-fed infants. This better absorption is because Correspondence to: A Robles Medina, Departamento de Ingeniería, Química, Universidad de Almería, Spain arobles@ual.es Contract/grant sponsor: Ministerios de Educación yculturaydecienciaytecnología, Spain; contract/grant number: 1FD and AGL Contract/grant sponsor: Plan Andaluz de Investigación; contract/grant number: CVI 0173 Received 28 May 2003; revised version received 4 June 2004; accepted 16 July 2004) Published online 12 October Society of Chemical Industry. J Chem Technol Biotechnol /2004/$

2 PA González Moreno et al pancreatic lipase hydrolyses ester bonds at the positions sn-1andsn-3intriacylglycerols, andshowshigher activity toward medium chain fatty acids than toward long chain fatty acids, especially polyunsaturated fatty acids PUFAs). 5 The liberated medium chain free fatty acids are directly absorbed into the portal vein and the 2-monoacylglycerols with the essential long chain fatty acids) are well absorbed via the lymphatic route. 6 The simplest and most direct route for the synthesis of ST of the MLM type is acidolysis between long chain triacylglycerols and medium chain free fatty acids catalysed with a 1,3-specific lipase The lipases offer high catalytic efficiency, specificity and selectivity by incorporation of the required acyl group into a specific position on the native triacylglycerol. Acyl migration is a major problem in the synthesis of STs in batch reactors and this decreases the yield in the target ST. The high substrate/enzyme ratios used in a typical reaction mean that a long time is required to attain equilibrium and this inevitably results in acyl migration. In contrast, an enzymatic reactor with the lipase immobilised in a packed column has advantages over the stirred batch reactor in reducing the acyl migration. 12 Packed-beds are commonly employed for solid fluid contacting in heterogeneous catalysis for several reasons: i) it facilitates the contact and subsequent separation between reactant and catalyst; ii) it allows reuse of the enzyme without the need for a prior separation; iii) a continuous mode of operation can be used easily; and iv) packed-bed are more costeffective than batch reactors. 9,11,12 In contrast to the wealth of information pertaining to the kinetics of hydrolysis and esterification reactions catalysed by lipases, little has been reported on the kinetics of lipase-catalysed interesterification between free fatty acids and heterogeneous triacylglycerols. Perhaps the most detailed kinetic study of this type is the work of Reyes and Hill. 13 These authors studied the kinetics of the acidolysis between free fatty acids and heterogeneous triacylglycerols such as olive oil and milk fat. They proposed a kinetic model that could account for the effect of the concentrations of all chemical species participating in the interesterification reaction. Also, little information has been reported on the magnitude of rate constants in the esterification reactions but this information is essential for rational design and scaling up of interesterification reactors. This work studies a method to produce structured triacylglycerols, using the immobilised 1,3-specific lipase Lipozyme IM as the catalyst. This lipase was immobilised in a packed-bed reactor and the reaction mixture was recirculated through the bed. Equilibrium and kinetics of ST s synthesis, the effects of the flow rate of the reaction mixture through the lipase bed and the substrate concentration on the conversion rate were investigated. A simple model was developed to account for the effects of enzyme load, reactant flow rate and the substrate concentration on the composition of synthesised ST. This model estimates kinetic constants for all the fatty acids present native and caprylic acid). 2 MATERIALS AND METHODS 2.1 Chemicals and materials Lipozyme IM was donated by Novo Nordisk A/S Bagsvaerd, Denmark). This lipase, contained 2 3% water as determined by Karl Fischer titration Compact titrator microkf 2026, Crimson, Alella, Spain), and was supplied immobilised on a macroporous anion-exchange resin. The enzyme showed a 1,3- positional specificity. Analytical grade caprylic acid CA) and hexane were obtained from Sigma Aldrich St Louis, MO). Table 1 shows the fatty acid composition of the commercial oil used EPAX4510, Pronova Biocare, Norway). From this composition an average molecular weight for the EPAX4510 was Da. 2.2 Acidolysis reaction in the packed-bed reactor PBR) Figure 1 shows a scheme of the reaction system. The immobilised lipase 2 3 g) was packed into a glass column 6.6id 250 mm length) covered with aluminium foil to prevent photo-induced oxidation. The enzyme bed was held between two mobile perforated disks. The porosity in this confined bed was 0.46 and the bed density of catalyst was 0.36 g cm The substrate mixture was kept in a reservoir submerged in a thermostatted water bath. The mixture consisted of EPAX4510, 10 or 20 g; caprylic acid, 9.8 g or 20 g; and hexane, cm 3. The molar ratio m 0 ) of caprylic acid/epax4510 was 6 in all the experiments. The initial EPAX4510 Table 1. Average values of the fatty acid composition of the original EPAX4510 F X0 ), the structured triacylglycerols at equilibrium F Xe ), the estimated fatty acid composition of the position sn-2 of triacylglycerols F X2 ) and equilibrium constants of exchange of different EPAX4510 native fatty acids with caprylic acid K i ) eqns 4) and 5)) Fatty acids F X0 % moles) F Xe % moles) F X2 a eqn 6)) % moles) F X2 b statistical distribution) % moles) K i 8: ± : ± : ± : ± :1n ± :1n ± :4n ± :1n ± :4n ± :5n ± :1n ± :6n ± Total a Calculated by eqn 6). b Calculated by eqns 2) and 3) when K i = K = J Chem Technol Biotechnol 80: )

3 Production of structured triacylglycerols C 4 Omegawax 0.25 mm 30 m, 0.20 µm standard film; Supelco, Bellefonte, PA), and a flame-ionisation detector. Nitrogen was the carrier gas and the total column flow was 36 cm 3 min 1. The oven temperature program was the following: 150 C for 3 min, from 150 C to 185 Cat10 Cmin 1, 185 Cfor10min, from 185 C to 240 Cat10 Cmin 1, and finally 240 C for 12 min. Matreya Pleasant Gap, PA) n-3 PUFAs standard catalogue number 1177) was used for the qualitative fatty acid determination. Caprylic acid was also injected to determinate its retention time. Nonadecanoic acid 19:0) was used as an internal standard for quantitative determination of fatty acids. The signal was analysed and integrated by an on-line computer. The masses of fatty acids were calculated by the equation: Figure 1. Immobilised lipase packed-bed reactor. 1) Substrate reservoir, 2) reactor temperature control, 3) peristaltic pump, 4) water jacket, 5) bed of immobilised lipase, 6) three-way valve, 7) sampling, 8) cooling/heating water. concentration was varied from M to M. In addition, some experiments were conducted without the solvent. The reaction mixture was pumped upward through the column by a peristaltic pump at flow rates between 30 and 200 cm 3 h 1. The column was jacketed to control the reaction temperature 30 C). The reactor was operated by recirculating the mixture leaving the bed to the substrate feed reservoir. The reaction was followed by sampling at different times between 1.0 and 175 h) from the substrate reservoir, which was continuously agitated at 200 rpm. The samples were stored at 20 C until analysis. All analyses were carried out in triplicate. The standard deviation was always below 6%. 2.3 Identification of reaction products and estimation of the molar fraction of fatty acids in triacylglycerols Hexane was removed from the product mixture in a vacuum evaporator. Glycerides monoacylglycerols, diacylglycerols and triacylglycerols) were extracted with 3 3cm 3 ) hexane after adding 2 cm 3 of 0.5 N KOH 20% ethanol solution) to 70 mg of the reaction mixture. These glycerides were identified by thinlayer chromatography TLC) followed by quantitative gas chromatography GC). TLC analysis has been described elsewhere. 15 Fractions corresponding to each glyceride type were scraped from the plates and methylated by direct transesterification with acetyl chloride/methanol 1:20) using the method of Lepage and Roy. 16 These methyl esters were analysed with a Hewlett Packard 4890 gas chromatograph Avondale, PA) connected to a capillary column of fused silica 3 Fatty acid mg) = f Xarea X 1) area IS where is the amount of the internal standard 19:0), f X is the fatty acid response factor, area X is the fatty acid chromatographic area and area IS is the internal standard chromatographic area. The response factors were close to 1 for all the fatty acids of molecular weight close to the internal standard and therefore a value of f X = 1 was taken for all the fatty acids, except for the caprylic acid. For the latter, f x was EQUILIBRIUM AND RATE EQUATIONS 3.1 Acidolysis equilibrium A general treatment of the equilibrium for the reactions of acidolysis catalysed by 1,3-specific lipases has been reported recently. 14 The treatment is valid for the acidolysis between any heterogeneous triacylglycerol in which participate n native fatty acids L i, i = 1..n), and an odd fatty acid M). It is based on the following hypotheses: i) due to the 1, 3 positional specificity of lipase, only the fatty acids in the positions sn- 1andsn-3 of triacylglycerol are exchanged; ii) the exchange in position sn-1 does not depend on the nature of the fatty acid in position sn-3 and vice versa; iii) the concentration of partial acylglycerols mono- and diacylglycerols) is negligible at any time. The latter was experimentally observed under the operational conditions used. The treatment leads to the equations: K 1 F Lie F Li2 ) )) = K i F Li0 F Lie ) 1 + K K i F Li0 F Lie F Me K = i F Me F Li0 F Lie ) K i ) + K )) 2) 3) J Chem Technol Biotechnol 80: ) 37

4 PA González Moreno et al which allow the calculation of the exchange equilibrium constant, K i, for each native fatty acid, L i, by the odd fatty acid, M. K is the average equilibrium constant for the exchange of the native fatty acid L by the odd fatty acid M; m 0 is the initial caprylic acid/triacylglycerol molar ratio; F Me is the molar fraction of caprylic acid incorporated into ST at equilibrium; and F Li0, F Lie and F Li2 are the molar fractions of a native oil fatty acid, L i, in the original oil triacylglycerols, in the ST at equilibrium and esterifying the sn-2 position of the ST at the equilibrium, respectively. All these molar fractions are determined by gas chromatography, except F Li2, which cannot easily be determined. If the nature of the native fatty acid does not affect its degree of exchange with the odd fatty acid, K = K 1 =...= K n = 1, a statistical distribution of the fatty acids present at positions sn-1 and sn-3 would be obtained when the equilibrium was attained, and the n equations 2) permit the calculation of the fatty acid composition of the position sn-2 F x2 for a statistical distribution of fatty acids, Table 1). When the odd fatty acid/native triacylglycerol molar ratio is sufficiently high the degree of incorporation of the odd fatty acid M at equilibrium is high and the concentration of native triacylglycerols must be negligible. Under those conditions the equilibrium constant can be determined by the simplified equations: F Me F Me 1.5FMe ) F Me 2 2 K = ) 4) K i = F L i0 F Lie ) F Lie F Li2 ) 1 + K 1 + 2K )) )) ) 5) F Li2 was estimated by considering its concentration to be proportional to the content of that fatty acid L i )at equilibrium, ie: F Li2 = 0.33 F L ie = 0.33 F Lie 6) F Lie 1 F Me i This equation is based on the fact that at the equilibrium the majority of the native fatty acids are placed in the position sn-2 of the glycerol backbone, because these native fatty acids have been displaced quasi-quantitatively from the positions sn-1 and sn-3. If so, and all the triacylglycerols at the equilibrium were of the MLM type, the denominator i F L ie = 1 F Me which represents the total molar fraction of native fatty acids) becomes 0.33 and F Li2 = F Lie. 3.2 Kinetics of acidolysis In the same work 14 an empirical kinetic model was proposed for obtaining representative values of the apparent kinetic constants, k X, for each fatty acid, X native, L i, or odd, M), by assuming that the rate of incorporation of a fatty acid X into triacylglycerols by a unit amount of enzyme, r X mol h 1 g 1 of lipase), is proportional to the extent of deviation from the equilibrium ie the driving force) for each fatty acid; thus, r X = k X F Xe F X ) 7) F X is the molar fraction of any fatty acid odd or native) in the triacylglycerols and F Xe is the molar fraction at equilibrium. r X is positive for caprylic acid M) and negative for the native oil fatty acids L i ). To calculate reaction rates when the reaction occurs in an immobilised lipase bed, we determined the average reaction rate r mx, mol h 1 g 1 ) between the entrance and the exit of the lipase bed. r mx can be obtained by means of a mass balance for any free fatty acid, X, applied to the bed; ie q[x] in r mx m L = q[x] out 8) where m L is the lipase mass in the bed g), q is the flow rate of the substrate mixture through the lipase bed cm 3 h 1 )and[x] in and [X] out are the concentrations mol cm 3 ) of the free fatty acid M or L i )atthe entrance and at the exit of the bed, respectively. When we operated with recirculation of the reaction mixture through the lipase bed, the operation was under non-steady state conditions and the accumulation term in the balance was neglected. When the volume of reaction mixture held in the feed reservoir was large compared with the volume in the bed, the change of [X] with time due to the accumulation was negligible compared with the change of [X] due to the enzymatic reaction in the bed. The relationships between the concentration of a free fatty acid caprylic acid, [X] = [M], or a native fatty acid, [X] = [L i ]) and its molar fractions in the triacylglycerols F X = F M or F X = F Li ) is given by corresponding balances as follows: m 0 [TG] 0 = [X] + 3[TG] 0 F X = [M] + 3[TG] 0 F M 9) 3[TG] 0 F Li0 = [X] + 3[TG] 0 F X = [L i ] + 3[TG] 0 F Li 10) These balances ignore partial acylglycerols mono- and diacylglycerols) because the concentrations of these acylglycerols were negligible in all the experiments we performed. By substituting eqn 9) for caprylic acid) and 10) for native fatty acid) into eqn 8), we obtain: r mx = 3[TG] 0F Xout F Xin ) 11) m L /q 38 J Chem Technol Biotechnol 80: )

5 Production of structured triacylglycerols In order to facilitate the interpretation of the experiments, it was necessary to work with a high circulation flow rate to eliminate the possible influence of the external mass transport effects in the bed. Also, the feed reservoir should be well mixed. To ensure that the latter condition was attained, the reaction mixture reservoir was agitated continuously. When the feed reservoir is perfectly mixed, a differential mass balance of any free fatty acid, X, in the feed reservoir provides: q[x] out dt = q[x] in dt + V d[x] 12) because the mixture leaving the bed is identical to that entering the reservoir and vice versa. In eqn 12), V is the volume of reaction mixture in the reservoir. This equation can be rewritten in the form: V d[x] ) = q[x] in [X] out ) 13) dt Comparison of eqns 8) and 13) shows that if the volume contained in the feeding reservoir is large compared with the volume contained in the outer flow circuit, we can conclude that: V d[x] ) = r mx 14) dt m L The first term of eqn 14) corresponds to the reaction rate r X ) in a batch perfectly mixed reactor. According to the proposed model, r X is given by eqn 7). Therefore, considering that r X = r mx and considering eqns 9) and 10), we have: r X = 3[TG] 0V m L df X dt 15) Equaling expressions 7) and 15) and integrating between time zero and a time t, we obtain: F X = F Xe + F X0 F Xe ) exp k ) Xm L t 16) 3[TG] 0 V Equation 16) represents the variation of the fatty acid composition of the structured triacylglycerols with time in a perfectly mixed discontinuous reactor. 4 RESULTS AND DISCUSSION 4.1 Equilibrium of acidolysis between EPAX4510 and caprylic acid Figure 2 5 show the variation with time of the incorporation of caprylic acid and EPA in structured triacylglycerols. Figures similar to Figs 3 and 5 were obtained for the other native fatty acids of EPAX4510. These experiments were performed at different flow rates of reaction mixture through the lipase bed Figs 2 and 3) and at different triacylglycerol and caprylic acid concentrations Figs 4 and 5). These figures show that, except for the experiments carried out at the highest concentration of EPAX4510 [TG] 0 = F M cm 3 h cm 3 h time, h Figure 2. Influence of time and reaction mixture flow rate on the mass fraction of caprylic acid F M ) in the structured triacylglycerols obtained by acidolysis of EPAX4510 and caprylic acid in a packed-bed reactor with recirculation. Acidolysis conditions: 10 g of EPAX4510, 10 g of C8:0, 300 cm 3 hexane [TG] 0 = M), 2.1 g of Lipozyme IM. The solid continuous lines were obtained by using eqn 16), the values of F X0 and F Xe of Table 1 and the initial concentration of triacylglycerols [TG] 0 ), lipase mass m L ) and hexane volume V) as indicated above. F EPA cm 3 h cm 3 h time, h Figure 3. Influence of time and reaction mixture flow rate on the mass fraction of EPA F EPA ) in the structured triacylglycerols obtained by acidolysis of EPAX4510 and caprylic acid in an immobilised lipase packed-bed reactor with recirculation. See caption of Fig 2 for the reaction conditions and the parameter used in obtaining the solid lines M) and without solvent, the equilibrium was attained at around 30 h. Figures 2 and 3 show that the composition of caprylic acid and EPA, respectively, at the equilibrium 59.5% and 9.6%, respectively, Table 1) were practically the same at the two flow rates tested 74 and 196 cm 3 h 1 ). Figures 4 and 5 show that the composition of caprylic acid and EPA at the equilibrium was practically the same at the two lower concentrations tested and M). When the concentration was higher M) and when no solvent was used equilibrium was not attained at the long reaction times tested Figs 5 and 6), however, it is likely that the equilibrium concentration achieved after longer reaction times would not much differ from those shown in Figs 4 and 5 at the initial concentrations of EPAX4510, and 0.108M. These results show J Chem Technol Biotechnol 80: ) 39

6 PA González Moreno et al F M [TG] 0 =0.036 M [TG] 0 =0.108 M [TG] 0 =0.216 M without solvent time, h Figure 4. Influence of time and triacylglycerols concentration on the mass fraction of caprylic acid F M ) in the structured triacylglycerols obtained by acidolysis of EPAX4510 and caprylic acid in an immobilised lipase packed-bed reactor with recirculation. Acidolysis conditions: g of EPAX4510, g of C8:0, cm 3 hexane, 2.5 g Lipozyme IM, 200 cm 3 h 1. Experiment without solvent: 20 g EPAX4510, 20 g C8:0, 0 cm 3 hexane, 3.0 g of Lipozyme IM, 30 cm 3 h 1. The solid continuous lines were obtained by using eqn 16), the values of F X0 and F Xe of Table 1 and the initial concentration of triacylglycerols [TG] 0 ), lipase mass m L ) and hexane volume V) previously indicated. F EPA [TG] 0 =0.036 M [TG] 0 =0.108 M [TG] 0 =0.216 M without solvent time, h Figure 5. Influence of time and triacylglycerols concentration on the mass fraction of EPA F EPA ) in the structured triacylglycerols obtained by acidolysis of EPAX4510 and caprylic acid in an immobilised lipase packed-bed reactor with recirculation. See caption of Fig 4 for the reaction conditions and the parameter used in obtaining the solid lines. that the composition of caprylic acid and EPA of the structured triacylglycerols at the equilibrium was independent of the reaction time, flow rate Figs 2 and 3) and the initial concentration of EPAX4510 for [TG] M Figs 4 and 5). Table 1 shows the mean equilibrium composition of the structured triacylglycerols. The mean equilibrium composition for caprylic acid F Me ) was 59.5%. This composition oscillated between 57.1% and 61.4% and never exceeded the theoretical 66.7%. This result and the claimed 1,3 positional specificity of Lipozyme IM corroborate that only positions sn-1 and sn-3 of triacylglycerols participate in the exchange. Caprylic acid incorporation at equilibrium 59.5%) was only slightly higher than that obtained using cod liver oil 57%). 14 The average EPA content at equilibrium was 9.6% Table 1). Table 1 also shows the equilibrium constants K i ) calculated from the average equilibrium compositions by using eqns 4) and 5). The K i values varied for the different fatty acids, which indicated that the nature of the fatty acid influenced its degree of displacement by caprylic acid. The saturated and monounsaturated fatty acids [except oleic acid 18:1n9)] had exchange constants less than one, which indicated that their displacement by caprylic acid was not favoured. It was also observed that the acids 18:1n7 and 18:1n9 had different equilibrium constants 0.44 and 1.51, respectively) although they have the same number of carbon atoms and double bonds. Thus, it seems that the position of the double bond also influences the equilibrium constant value. The PUFAs generally had equilibrium constant values appreciably greater than one, which indicates that they were the acids most exchanged and that their displacement by caprylic acid was most favoured. Table 1 also shows the fatty acid composition that occurs at position sn-2, F Li2. These values have been calculated in two ways: by eqn 6) and also by eqns 2) and 3), assuming a statistical distribution of the fatty acid in the positions sn-1 and sn-3. If the distribution is statistical, the nature of the native fatty acid does not affect their degree of exchange with the odd fatty acid in the positions sn-1 and sn-3 and the tendency to occupy the positions sn-1 and sn-3 of the glycerol backbone is the same for all the native fatty acids and for the odd fatty acids, ie K = K 1 =...= K i = 1. In this case the fatty acid composition in the positions sn-1 and sn-3 of the structured triacylglycerols at the equilibrium would depend only on the initial concentration of each fatty acid, the natives at the positions sn-1 and sn-3 of the initial triacylglycerol and the odd as free fatty acid. F X2 values show that the compositions obtained by both methods are similar and that the fatty acids that preferentially occupy the position sn-2 are EPA, oleic acid 18:1n9), 18:1n7 and palmitic acid 16:0). This result is logical for EPA because the original oil contained 40% of EPA. Nevertheless, 8 or 9% EPA in position sn-2 of the structured triacylglycerol is a small percentage and it seems to indicate that most of the EPA contained in the original oil occupied sn-1 and sn-3 positions. In this sense, if we assume a random distribution of fatty acids in the starting material, the final structured triacylglycerol should contain a maximum of 13.2% EPA, taking into account the 1 3 positional specificity of the lipase. According to Table 1, only 19 24% of the original EPA of EPAX4510 was in position sn-2. This percentage is insufficient to synthesise structured lipids of MLM type with a high percentage of EPA occurring in position sn-2. This relatively small proportion of PUFAs in the position sn-2 contrasts with those referenced by Xu et al 11 for a refined fish oil, Shimada 40 J Chem Technol Biotechnol 80: )

7 Production of structured triacylglycerols et al 8 for tuna oil and Camacho Páez et al 14 for cod liver oil. To make a more cost-effective process, the free fatty acids, highly rich in EPA, could be recovered from the reaction products, by adding a base KOH or NaOH) and extracting the triacylglycerols with an organic solvent hexane). 17 The caprylic acid could then be separated from the long chain fatty acids highly rich in EPA) by step-wise chromatographic elution. 17,18 This result could also be due to the existence of some degree of acyl migration of EPA into the triacylglycerol from the position sn-2 to the more exchangeable positions sn-1 andsn-3. Thisacylmigrationmayoccur because the support on which the lipase is immobilised is an anion-exchange resin that catalyses this process. 19 The acyl migration, if any, must be through the mono- and diacylglycerols. 20 The concentration of these partial acylglycerols is almost negligible, due to the almost total absence of water. Nevertheless, a small amount of diacylglycerols must have formed, because these products are essential intermediates in the acidolysis. 21 This hypothesis is also supported by results obtained in our laboratory when the operation was carried out in the same system operating in continuous mode. 22 In this case the time of contact between the lipase and the reaction mixture was much less and the EPA content of the structured triacylglycerol at the equilibrium was 20%. Therefore, in order to avoid the acyl migration, the time of contact between the triacylglycerol and Lipozyme IM should be kept to a minimum or a different type of support should be used to immobilise the enzyme. 4.2 Kinetics of the acidolysis of EPAX4510 and caprylic acid First, the influence of the reaction mixture flow rate on reaction rate was studied with the purpose of operating with mixture flow rates that eliminated the possible influence of the external mass transfer. Figures 2 and 3 show that the incorporation rate of caprylic acid and the displacement rate of EPA at the flow rates of 74 and 196 cm 3 h 1 were practically coincident, indicating that in the conditions used the acidolysis, rate was independent of the flow rate. Next, experiments in which the reaction mixtures flow rate was maintained constant and the triacylglycerol and caprylic acid concentrations were progressively increased maintaining the molar ratio of the two substrates constant) were carried out. Figures 4 and 5 show that for the range of concentrations tested, the concentration had a strong influence on the acidolysis rate. This influence was particularly significant for the experiment performed at the concentration of M and when solvent was not used. All these results have been adjusted to the model represented by eqn 7). As previously mentioned, when the acidolysis rate is independent of the flow rate, the reaction system behaves as a perfectly mixed batch reactor and the experimental results should fit eqn 16). Thus the experimental results shown in Figs 2 5, corresponding to caprylic acid and EPA, were fitted with eqn 16). Table 2 shows the values of the kinetic constants, k X, that best fitted the experimental results. Figures 2 5 show the model fitting for the caprylic acid and EPA for both the flow rates tested Figs 2 and 3) and the two smallest triacylglycerol concentrations tested Figs 4 and 5). In these cases the mean deviation found was never more than 10%, which is small considering the manipulations that are necessary for the analysis of the samples. For the experiments carried out at the highest concentration 0.216M) and without solvent the error was somewhat greater. As noted above, in this case the deviations could also be due to the influence of the mass transfer on the kinetics of the process and, therefore, to deviation of the behaviour of the system from a perfectly mixed reactor. As depicted in Figs 2 and 3 and Table 2, the flow rate had practically no influence on the reaction rate for the exchange of caprylic acid and EPA. This result shows that the reaction rate is not influenced by the external mass transfer to the catalyst particles. Figures 4 and 5 and Table 2 show that the higher the initial triacylglycerol concentration the lower the reaction rate was. In these cases it is likely that intraparticle diffusion effects 23 may explain the results. When the concentration of triacylglycerol was M, almost 40% of the mass of the reaction mixture was caprylic acid and triacylglycerols and 100% when no solvent was used. Taking into account the viscosities for hexane and caprylic acid mpa s and mpa s at 25 C, respectively), the viscosity of the reaction mixture in these two experiments was significant, and according to the Wilke and Chang equation, a concomitant drop of the effective diffusion coefficient of the substrates into the catalyst particles would take place. 24 Lipozyme IM is a granular product with a relatively large particle size of mm data of Novo Nordisk) and a relatively small average pore Table 2. Apparent kinetic constants k X,molg 1 h 1 ) that fit the experimental results for the exchange of caprylic acid and EPA to different reaction mixture flow rates through the lipase bed and different initial concentrations of triacylglycerols Flow rate q) and initial triacylglycerol concentration, [TG] 0 Caprylic acid 8:0) k X mol g 1 h 1 ) Eicosapentaenoic acid 20:5n-3, EPA) q = 74 cm 3 h 1a q = 196 cm 3 h 1a [TG] 0 = M b [TG] 0 = M b [TG] 0 = M b Without solvent c a Experiments carried out at [TG] 0 = M see caption of Fig 2 for the reaction conditions). b Experiments carried out at 200 cm 3 h 1 see caption of Fig 4 for the reaction conditions). c Experiments carried out at 30 cm 3 h 1 see caption of Fig 4 for the reaction conditions). J Chem Technol Biotechnol 80: ) 41

8 PA González Moreno et al diameter of 260 Å. 25 With these characteristics, Ison et al 23 demonstrated that some internal mass transfer limitations may occur. In addition, the experiment without solvent was performed at a flow rate of only 30 cm 3 h 1 and, in this case, the external mass transfer could have also been influential. 5 CONCLUSIONS Equation 16) can be used to predict the time variation of the composition of the structured lipids in a packed-bed reactor operated in batch recirculation. Also this equation could be useful to scale up this reactor. Predictions using eqn 16) agreed with the measurements at the smallest concentrations of triacylglycerol tested and M) and in the range of flow rates, cm 3 h 1, used. Scaling up of this reaction could be carried out by maintaining the parameter m L t/v [TG] 0 constant eqn 16)). This parameter represents the intensity of enzymatic treatment. In scaling up it is also important to keep all the variables that affect the reaction rate constant, such as the initial triacylglycerol and caprylic acid concentrations. The fluid dynamics should be selected to minimise the influence of solid liquid mass transfer on the reaction. The results obtained with the acidolysis carried out in a packed-bed reactor operating in batch mode show that this method is very useful for producing structured triacylglycerols of the MLM type incorporating significant amounts of caprylic acid 59.5%) and with 9.6% of EPA in the position sn-2. Although better absorption of the MLM type structured triacylglycerol with respect to a random fatty distribution of PUFAs has been proved, 3,4 the content of EPA of the final ST obtained in batch mode operation is still low 9.6%) taking into account that the original oil contains 40% of this PUFA. ACKNOWLEDGEMENTS We thank Y Chisti for his help in the English edition of the paper. This research was supported by grants from the Ministerio de Educación y Cultura and Ministerio de Ciencia y Tecnología Spain), Projects 1FD and AGL , and Plan Andaluz de Investigación, CVI REFERENCES 1 Xu X, Hoy C-E, Blachen S and Ander-Nissen J, Specificstructured lipid: nutritional perspectives and production potentials, in Proceedings of International Symposium on the Approaches to Functional Cereals and Oils. CCOA, Beijing, pp ). 2 Xu X, Production of specific-structured triacylglycerols by lipase-catalyzed reactions: a review. Eur J Lipid Sci Technol: ). 3 Christensen MS, Hoy CE, Becker CC and Redgrave TG, Intestinal absorption and lymphatic transport of eicosapentaenoic EPA), docosahexaenoic DHA), and decanoic acids: dependence on intramolecular triacylglycerol structure. Am J Nutr 61: ). 4 Kennedy K, Fewtrell MS, Morley R, Abbott R, Quinlan PT, Wells JCK, Bindels JG and Lucas A, Double-blind, randomized trial of a synthetic triacylglycerol in formula-fed term infants: effects on stool biochemistry, stool characteristics, and bone mineralization. Am J Clin Nutr 70: ). 5 Bottino NR, Vandenburg GA and Reiser R, Resistance of certain long-chain polyunsaturated fatty acids of marine oils to pancreatic lipase hydrolysis. Lipids 2: ). 6 Jandacek RJ, Whiteside JA, Holcombe BN, Volpenheim RA and Taulbee JD, The rapid hydrolysis and efficient absorption of triglycerides with octanoic acid in the 1 and 3 positions and long-chain fatty acid in the 2 position. Am J Clin Nutr 45: ). 7 Shimada Y, Sugihara A, Nakano H, Yokota T, Nagao T, Komemushi S and Tominaga Y, Production of structured lipids containing essential fatty acids by immobilized Rhizopus delemar lipase. J Am Oil Chem Soc 73: ). 8 Shimada Y, Sugihara A, Maruyama K, Nagao T, Nakayama S, Nakano H and Tominaga Y, Production of structured lipid containing docosahexaenoic and caprylic acids using immobilized Rhizopus delemar lipase. J Ferment Bioeng 81: ). 9 Shimada Y, Suenaga M, Sugihara A, Nakai S and Tominaga Y, Continuous production of structured lipid containing γ - linolenic and caprylic acids by immobilised Rhizopus delemar lipase. J Am Oil Chem Soc 76: ). 10 Akoh CC and Huang KH, Enzymatic synthesis of structured lipids: transesterification of triolein and caprylic acid. J Food Lipids 2: ). 11 Xu X, Balchen S, Høy CE and Adler-Nissen J, Pilot batch production of specific-structured lipids by lipase-catalyzed interesterification: preliminary study on incorporation and acyl migration. J Am Oil Chem Soc 75: ). 12 Mu H, Xu X and Høy C-E, Production of specific-structured triacylglycerols by lipase-catalyzed interesterification in a laboratory-scale continuous reactor. J Am Oil Chem Soc 75: ). 13 Reyes HR and Hill CG, Kinetic modeling of interesterification reactions catalyzed by immobilized lipase. Biotechnol Bioeng 43: ). 14 Camacho Páez B, Robles Medina A, Camacho Rubio F, González Moreno P and Molina Grima E, Production of structured triglycerides rich in n-3 polyunsaturated fatty acids by the acidolysis of cod liver oil and caprylic acid in a packed-bed reactor: equilibrium and kinetics. Chem Eng Sci 57: ). 15 Robles Medina A, Esteban Cerdán L, Giménez Giménez A, Camacho Páez B, Ibáñez González MJ and Molina Grima E, Lipase-catalyzed esterification of glycerol and polyunsaturated fatty acids from fish and microalgae oils. J Biotechnol 70: ). 16 Lepage G and Roy C, Improved recovery of fatty acid through direct transesterification without prior extraction or purification. J Lipid Res 25: ). 17 Robles Medina A, Giménez Giménez A, García Camacho F, Sánchez Pérez JA, Molina Grima E and Contreras Gómez A, Concentration and purification of stearidonic, eicosapentaenoic, and docosahexaenoic acids from cod liver oil and themarinemicroalgaisochrysis galbana. J Am Oil Chem Soc 72: ). 18 Belarbi H, Molina E and Chisti Y, A process for high yield and scaleable recovery of high purity eicosapentaenoic acid esters from microalgae and fish oil. Process Biochemistry 35: ). 19 Millqvist Fureby A, Virto C, Adlercreutz P and Mattiasson B, Acyl group migration in 2-monoolein. Biocatal Biotrans 14: ). 20 Serdarevich B, Glyceride isomerization in lipid chemistry. JAm Oil Chem Soc 44: ). 21 Ainsworth S, Versteeg C, Plamer M and Millikan MB, Enzymatic interesterification of fats. Milkfat Update Conference, 42 J Chem Technol Biotechnol 80: )

9 Production of structured triacylglycerols February 1996, Werribee, Victoria Australia). Australian J Dairy Technol 51: ). 22 González Moreno PA, Robles Medina A, Camacho Rubio F, Camacho Páez B and Molina Grima E, Production of structured lipids by acidolysis of an EPA enriched fish oil and caprylic acid in a packed bed reactor: analysis of three different operation modes. Biotechnol Prog 20: ). 23 Ison AP, Macrae AR, Smith CG and Bosley J, Mass transfer effects in solvent free fat interesterification reactions: influences on catalyst design. Biotechnol Bioeng 43: ). 24 Perry RH and Chilton CH eds), Chemical Engineers Handbook, 5th edn. McGraw-Hill, New York, pp ). 25 Camacho Páez B, Obtención de lípidos estructurados por acidolisis con lipasas inmovilizadas. Doctoral thesis, Departamento de Ingeniería Química, Universidad de Almería 2000). J Chem Technol Biotechnol 80: ) 43